In this article, we examined the achievement composition effect on the scientific and computer literacy from sixth to ninth grade of 5,738 German students in secondary school. We used multilevel modeling in a structural equation modeling framework to estimate doubly latent models, which control for measurement error and sampling error. Results indicated that class-average competence was positively related to students' individual competence in secondary school. Composition effects remained significant for computer literacy but not for scientific literacy when further control for potential selection bias was added by accounting for demographic and background differences between students. When taking school track (academic vs. vocational) into account to separate institutional effects from peer spillover effects, results showed, on the one hand, that there was no positive peer spillover effect for computer literacy left when controlling for school track. We even found a negative effect on scientific literacy, which means that students reached a higher scientific literacy when they came from lower-achieving classes. On the other hand, we found strong effects from school track on both competencies in favor of students from academic-track schools. This indicates that the overall achievement composition effect can be primarily traced back to tracking effects. This finding suggests that vocational-track schools may need teacher professional development programs to increase teaching quality.
